Insights / Virtual power plants · Aug 2026
Cloudline: can a virtual power plant steer batteries it never installed?
A market study for Flexa on reaching Germany's existing home batteries through the manufacturers' clouds instead of new hardware: how big the market gets by 2030, which offer wins customers, whether the unit economics hold, and what to test before investing.
- Serviceable homes 2030
- 1.33m
- Realistic reach 2030
- 213k homes
- Contribution per home
- €61/yr
Flexa runs one of Europe’s largest residential virtual power plants: it connects home batteries, trades their flexibility on power markets and shares the value with households. Today every Flexa home needs an IoT box, fitted by an installer when the solar system is sold, so Flexa only reaches a household at the moment of a new installation.
Germany already has about 2.8 million home batteries in operation (MaStR via the Flexa battery explorer, Oct 2026). This study looks at a different route to them: steering batteries through their manufacturers’ clouds, via a platform that has already integrated the major brands. No box and no installer; onboarding becomes a “connect your account” flow sold directly to consumers. I call the play Cloudline.
Two starting assumptions frame it. Cloud steering is less precise than a box, so a cloud-connected home earns about €250 a year of flexibility value, roughly half of a box-connected one. The integration platform charges €2 per home per month. Everything else below comes from my own market model, with each assumption labelled as one.
How big the market gets
Not every battery can be reached. Its brand must be integrated with the platform, the home needs a smart meter to trade flexibility properly, and the battery must be large enough to earn meaningful value. Step through the years to see the serviceable market widen as integrations, smart meters and battery sizes grow.
The fleet grows from 2.8 to about 4.9 million batteries by 2030, but the combined filter matters more: addressability widens from 10% to 27%, mostly because the smart-meter rollout catches up. The serviceable market grows from 292,000 homes in 2026 to 1.33 million in 2030.
What Flexa can realistically win
No clean benchmark exists for winning existing battery owners directly. Comparable programmes, such as 1KOMMA5° Heartbeat, sonnenCommunity and Flexa’s own box channel, all grew from their own installation base. I anchored the first year at 20,000 homes, a fifth of what Flexa reached through its captive channel, and let penetration flatten as the early adopters with large, compatible batteries are used up.
Connected homes, three adoption paths
Homes connected through the manufacturers' clouds at year end
- Base case
- Bear to bull range
Year end (2026 is the build year)
Data table
| Year end (2026 is the build year) | Base case | Bull case | Bear case |
|---|---|---|---|
| 2030 | 213k | 306k | 126k |
| 2029 | 135k | 174k | 74.8k |
| 2028 | 71.3k | 99.8k | 35.7k |
| 2027 | 20k | 28.7k | 9.6k |
| 2026 | 0 | 0 | 0 |
Penetration of the serviceable market in 2030: 9.5% bear, 16% base, 23% bull. At 10.3 kWh per home, the base case adds up to 2.2 GWh of steerable storage.
Source Cloudline market model (Aug 2026); penetration ramps are assumptions
The range is wide on purpose: 126,000 to 306,000 homes by 2030, with 213,000 in the base case. Which path materialises is exactly what a short, cheap test should reveal before any large commitment.
The offer that wins customers
Framing matters more than the amount. For a typical home with 10 kWp of solar and a 10 kWh battery, returning €100 a year as a higher feed-in rate moves it by only 1.6 ct per kWh. Returning the same €100 as a lower purchase price cuts it by 7.4 ct per kWh, because the home buys far less power than it exports. Same cash, 4.6 times the headline movement.
I compared four offers at the same €250 of value per home. Each returns a different share to the customer, costs a different amount to serve, and implies a different churn rate.
The winner is certainty: power at 28 instead of 35 ct per kWh and a feed-in rate that cannot fall, on a 12-month contract. It returns less cash than the others but promises something nobody else does, which should keep churn lowest.
Unit economics per home
At €61 contribution, 14% churn and €180 to acquire a customer, lifetime value covers acquisition cost 2.4 times, below the 3x a healthy subscription business needs. Three single levers each close the gap: €300 of value per home, a €90 give-back, or acquisition at €145. A cheaper platform fee comes close: at €1 per month instead of €2, the ratio rises to 2.9x.
The cash it takes
Growth burns cash before it earns it
Cumulative cash flow, € million
- Base case
- Bear to bull range
Year
Data table
| Year | Base case | Bear case | Bull case |
|---|---|---|---|
| 2030 | −€35.1m | −€25.9m | −€46.2m |
| 2029 | −€24.2m | −€16.9m | −€28.8m |
| 2028 | −€14.2m | −€9.3m | −€18.2m |
| 2027 | −€5.3m | −€3.7m | −€6.6m |
| 2026 | −€0.8m | −€0.8m | −€0.8m |
Faster growth digs a deeper hole: acquisition is paid up front and earned back over seven years. Contribution reaches about €11 million a year by 2030 in the base case.
Source Cloudline market model (Aug 2026), €180 acquisition cost
Every scenario stays cash-negative through 2030, and the bull case burns the most. The base case needs a cash commitment of about €35 million. It turns cash-positive only once growth slows, so acquisition cost decides both the size of the hole and how fast the business climbs out of it.
What has to be true before investing
I framed the decision as a gate with seven criteria, each measurable within 60 days and each with a pass and a kill threshold:
- Coverage: the platform reaches at least 55% of the fleet by brand, checked against the MaStR brand mix
- Reliability: at least 95% of steering commands succeed in a live pilot
- Value: at least €250 a year per home against an unsteered control group
- Acquisition cost: €145 or less in a paid test
- Acceptance: the offer converts at a lower give-back
- Overlap: little cannibalisation of the existing sales channel
- Mandate: the founders commit the cash
Any of the first four failing badly ends it. Otherwise the measured values go back into the model: go at 2.5x LTV/CAC or above, not yet between 2.0x and 2.5x, no below 2.0x. Two cheap desk audits run first, then a 20-to-40-home pilot and a small paid acquisition test run in parallel, for a total budget of about €20,000.
Why this belongs on EnergyMap
Most of the work went into stitching inputs together: building stock from Destatis, installations and smart-meter rollout from BSW-Solar and the Bundesnetzagentur, the brand mix from the Marktstammdatenregister, and tariff components from market sources. Anyone sizing a flexibility market rebuilds the same funnel. That is why a market funnel builder is next on the EnergyMap roadmap: official stock data, a source on every filter, and assumptions you can change.